At the Meta Connect conference on Oct. 6, 2026, Meta unveiled a new open standard called the personal agent protocol. The protocol is being co‑authored by Meta, Walmart, Stripe, the AI startup Sierra and a few other firms. Its goal is to give businesses a clear way to identify when an AI bot, like Meta’s Muse, is acting on a user’s behalf and to control what data the bot can access.
How the protocol works
The protocol works like the “Login with Google” button many sites already use. When a personal agent talks to a website, it will present a verified credential that tells the site the request comes from an AI acting for a specific person. The standard also includes logs that let businesses see what the agent is doing, similar to email headers. Developers say this should reduce the “chaos” of bots scraping sites or accessing credit‑card info without permission.
Industry reaction
Meta’s Muse, released in early September, quickly rose to the top of Apple’s App Store, overtaking ChatGPT. Walmart and Stripe see the protocol as a way to let customers safely share payment details with agents. Amazon, however, has blocked Muse and sued another AI agent maker over unauthorized web‑scraping. OpenAI and Anthropic are not yet part of the effort, but the group’s leader, former Salesforce co‑CEO Bret Taylor, said he hopes they will join.
Why it matters
For everyday shoppers, the protocol could let AI assistants book appointments or buy gifts without exposing personal data to unknown bots. For businesses, it offers a way to verify that an AI, not a human, is accessing their systems, reducing fraud risk. It remains unclear how quickly other AI providers will adopt the standard.
